問16
頻出度★★☆
教師あり学習において, 正解となる情報を付与する作業を表す用語として, 最も適切なものはどれか。
令和8年度 ITパスポート試験 問16
テクノロジ系基礎理論 › 情報に関する理論
This question's explanation and your answers so far.
解説
問 16
- ア
正解 Annotation is the term for the task of attaching correct-answer information (labels, tags) to data. Supervised learning is a method that trains a model by feeding it large numbers of pairs of an input data item and its correct answer (label), so tasks such as assigning the name of the object shown in an image or assigning a classification category to a sentence, that is, the act of attaching correct answers itself, are exactly annotation. The most appropriate term for the question's task of attaching correct-answer information is ア.
Incorrectイ
Encode is the task of converting data into another format or code (for example, converting characters into character codes, or converting category values into numbers). It only changes how the data is represented; it is not the task of attaching correct-answer information.
Incorrectウ
Data cleansing is preprocessing that removes or corrects errors, duplicates, inconsistent notation, missing values, and the like in the data to tidy it up. It is the task of making the data clean, not the task of attaching correct-answer labels.
Incorrectエ
Filtering is the task of extracting data that matches a condition or excluding unnecessary data. It is selecting which data to keep or discard, not the task of attaching correct-answer information.
ここがポイント
- Supervised learning needs training data with labels (correct answers), and the task of attaching correct-answer information to data is called annotation.
- Distinguishing the preprocessing terms: annotation (attaching correct-answer labels) / encode (encoding, format conversion) / data cleansing (cleaning up the data) / filtering (extracting or excluding by condition).
用語